Determining whether text-based content should be provided as a result to a query is typically a straightforward activity. For example, the text of the query can be matched against text appearing in the content and a text match score can be determined. If the score is high enough, the result is returned. Unfortunately, while techniques such as tagging and metadata can be used to describe non-text content, it is still typically more difficult for users to locate non-text content in which they are interested than text-based content.
Additionally, content providers, such as image hosting sites which allow users to upload and share photographs, are increasingly offering users ways of embedding dynamic content in their pages, such as by providing a JavaScript-based photo gallery widget. Some widgets, such as a quote of the day display, provide dynamic content (e.g., a different quote each day) but do not allow user interaction. Other widgets, however, such as the above mentioned gallery widget, might accept input from a user, such as a search query. Unfortunately, it can be difficult to determine what dynamic content should be returned to such a user.